Research map / Video / Video segmentation and tracking
Motion Sensitive Contrastive Learning for Self-Supervised Video Representation
Jingcheng Ni, Nan Zhou, Jie Qin, Qian Wu, Junqi Liu, Boxun Li, Di Huang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add L_RF cross-modality loss (ablation) | MSCL · UCF101 finetune | top-1 accuracy: 65.1 -> 71.4 | Improved |
| add L_LMC local motion loss (ablation) | MSCL · UCF101 finetune | top-1 accuracy: 65.1 -> 70.5 | Improved |
| add L_LMC on top of L_RF (ablation) | MSCL · UCF101 finetune | top-1 accuracy: 71.4 -> 75.6 | Improved |
| add L_LMC on top of L_RF (ablation) | MSCL · UCF101 retrieval R@1 | R@1: 36.5 -> 45.7 | Improved |
| add flow rotation augmentation (FA) (ablation) | MSCL · UCF101 finetune | top-1 accuracy: 75.6 -> 76.8 | Improved |
| add motion differential sampling (MDS) (ablation) | MSCL · UCF101 finetune | top-1 accuracy: 75.6 -> 76.4 | Improved |
| combine FA and MDS (ablation) | MSCL · UCF101 finetune | top-1 accuracy: 75.6 -> 77.3 | Improved |
| use 3D backbone in flow pathway when only L_RF is used (ablation) | MSCL · UCF101 finetune | top-1 accuracy: 71.4 -> 71.2 | No clear effect |
| use 2D ResNet-18 instead of R3D-18 for the flow pathway (ablation) | MSCL · UCF101 finetune | top-1 accuracy: 71.3 -> 72.3 | Improved |
| add reflect padding for 3D flow backbone instead of zero padding (ablation) | MSCL · UCF101 finetune | top-1 accuracy: 71.3 -> 70.7 | Hurt |
| add TPN in RGB pathway (ablation) | MSCL · UCF101 finetune | top-1 accuracy: 72.3 -> 75.6 | Improved |
| use flow input instead of motion boundary for L_LMC (ablation) | MSCL · UCF101 finetune | top-1 accuracy: 68.2 -> 70.5 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- RSPNet — RSPNet: Relative Speed Perception for Unsupervised Video Representation Learning (AAAI 2021) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Motion-Focused Contrastive Learning of Video Representations (ICCV 2021 · similarity 0.92)
- Compressed Video Contrastive Learning (NeurIPS 2021 · similarity 0.89)
- Probabilistic Representations for Video Contrastive Learning (CVPR 2022 · similarity 0.87)
- Spatiotemporal Contrastive Video Representation Learning (CVPR 2021 · similarity 0.87)
- MaCLR: Motion-Aware Contrastive Learning of Representations for Videos (ECCV 2022 · similarity 0.87)
- Self-Supervised Video Representation Learning in a Heuristic Decoupled Perspective (IJCV 2026 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
